Abstract Contemporary system identification algorithms are well proven to provide accurate eigenfrequency estimates in analyzing the systems with low modal damping. Since most engineering structures usually have low damping ratios the corresponding response characteristics can easily be obtained by conventional methods. Indeed, these modes can be extracted by using relatively short measurement durations (150–200 cycles of the lowest frequency included in the data block). However, some specific applications such as analyzing the in-operation vibration behavior of MW scale large wind turbines also require an accurate estimation of the modes with high damping. For a rotating wind turbine, some important turbine modes (e.g. flapwise rotor modes) have very high aeroelastic damping, which make them very difficult (if not impossible) to be detected. Extracting these high damping modes is a challenging task for almost all system identification techniques that are currently in use. In this work, a new method, which is based on Natural Excitation Technique (NExT), is proposed as an alternative approach for extracting the eigenfrequencies of high damping modes in an efficient way. NExT is a well established experimental dynamic analysis tool which was specifically developed to extract the dynamic characteristics of wind turbines in the early 90s. However, during the analyses it was observed that conventional NExT algorithm requires analyzing very long measurement durations (4500–5000 cycles) to be able to estimate the high damping modes accurately. A new method proposed in this work enables the eigenfrequencies of high damping modes to be estimated by using data series which are approximately 30 times shorter (around 150 cycles) than those required for a standard NExT algorithm.
«
Abstract Contemporary system identification algorithms are well proven to provide accurate eigenfrequency estimates in analyzing the systems with low modal damping. Since most engineering structures usually have low damping ratios the corresponding response characteristics can easily be obtained by conventional methods. Indeed, these modes can be extracted by using relatively short measurement durations (150–200 cycles of the lowest frequency included in the data block). However, some specific a...
»
Login
BibTeX